Hello,

As you may have seen ChEMBL team just release a new version of the RESTful webserive (http://chembl.blogspot.com/search?updated-max=2012-02-29T17:53:00Z&max-results=10), with a new possibility to search for similarity and substructure.

I tried to use this service in Pipeline, but it was not as smooth as I thought.

JSON format is not working correctly in PP (may be too many subcategories), and smiles can have symbols that doesn't work in URL.

S I decided to share what I ahd to do, in case some of you have the same problem, or have a better way than me to do it.

Attached are some examples I made for similarity and substructure search (with or output results only -with every properties, or output reference + compounds found with few properties).

I has some work around for the smiles, as strereochemistry with / and \ are removed, as well as + charge (on the smiles string).

Also it gets data using html reader to get data from ChEMBL, does some string replacement on the json data, and then use data from json to display correctly the data.

And I added a html reader to test the availibilty of the EBI webservice using the url:http://www.ebi.ac.uk/chemblws/status/

returning up if it is working.

What you need to run it is:

Data from JSON (not sure if it is by default or not)

Connection to the EBI website (we are using a proxy here in Dundee, otherwise our firewall wouldn't let me do it).

Tested on PP 8.5.0

The protocol contains 4 pipes for different scenarios, but it should be possible to have only 2, or even 1.

Cheers,

Jérémy